\begin{subtable}[p]{\linewidth} \vspace{2ex}\   
 

\caption*{Panel C: Causal Mediation Effects \hspace*{\fill}} \vspace{-2ex}
\begin{tabular*}{\hsize}{lcccccc}

\midrule
 \shortstack[c]{Outcome \vspace{1ex}} &\multicolumn{2}{c}{\shortstack[c]{Attended village council\\ meeting last year}} &\multicolumn{2}{c}{\shortstack[c]{Discusses politics and \\ community with husband}} &\multicolumn{2}{c}{\shortstack[c]{Participates in a \\ political organization}}\\ \cmidrule(lr){2-3}\cmidrule(lr){4-5}\cmidrule(lr){6-7} &\multicolumn{1}{c}{(1)} & \multicolumn{1}{c}{(2)} & \multicolumn{1}{c}{(3)} & \multicolumn{1}{c}{(4)}          &\multicolumn{1}{c}{(5)}  & \multicolumn{1}{c}{(6)}         \\ \shortstack[c]{Mediator \vspace{3ex}} & \shortstack[c]{Years \\education} & \shortstack[c]{No. of children \\ living with \\ respondent}  & \shortstack[c]{Years \\education} & \shortstack[c]{No. of children \\ living with \\ respondent}  & \shortstack[c]{Years \\education} & \shortstack[c]{No. of children \\ living with \\ respondent} \\
\midrule 
ACME &0.0004***&-0.0001***&0.0008***&-0.0003***&0.0001***&0.0000  \\ 
&(0.0002, 0.0005)&(-0.0002, -0.0000)&(0.0005, 0.0011)&(-0.0004, -0.0001)&(0.0000, 0.0001)&(-0.0000, 0.0000)  \\ 
ADE&0.0032**&0.0036**&0.0040*&0.0051**&0.0004&0.0004  \\ 
&(0.0004, 0.0061)&(0.0008, 0.0066)&(-0.0004, 0.0087)&(0.0007, 0.0097)&(-0.0005, 0.0013)&(-0.0005, 0.0014)  \\ 
ATE&0.0035**&0.0035**&0.0048**&0.0048**&0.0004&0.0004  \\ 
&(0.0008, 0.0065)&(0.0007, 0.0066)&(0.0005, 0.0094)&(0.0004, 0.0095)&(-0.0004, 0.0014)&(-0.0004, 0.0014)  \\ 
Proportion of ATE Mediated&0.1027&-0.0274&0.1580&-0.0527&0.0986&0.0102  \\ 
\bottomrule  

\end{tabular*}
\end{subtable}
\vspace{-1ex}
\captionsetup{justification=justified, width=\hsize}
\caption*{\footnotesize \textit{Notes:}
This table presents results for causal mediation analysis following the framework in Imai et al. (2011).
The mediators considered are \textit{Years education} and \textit{No. of children living with respondent}.
The treatment variable considered is \textit{Menarche age}.
Panel A reports the estimates for the OLS regression of the mediator on menarche age.
Panel B reports the estimates for the OLS regression of outcomes on menarche age and the mediator.
All regressions include district FEs, height, age, parents' education, household caste, religion. 
Panel C reports the estimates of the Average Causal Mediation Effect (ACME) and Average Direct Effect (ADE).
The Average Treatment Effect (ATE) corresponds to the results for the effect of menarche age on outcomes in Appendix Table \ref{table-effect-of-menarche-age-on-political-participation}.
The proportion of ATE Mediated is calculated as ACME divided by ATE.
In Panels A and B, robust standard errors are shown below coefficient estimates.
In Panel C, the 95\% confidence interval is shown below the estimate.
Data from IHDS 2011--12, rural women.
***$\,p < 0.01$, **$\,p < 0.05$, *$\,p<0.10$. \looseness=-1}
\end{table}
